Huixin Yu 1, Bernhard Steiert 1, Stefan Sturm 1
1 F. Hoffmann-La Roche Ltd (Basel, Switzerland)
Introduction: Non-linear pharmacokinetics (PK) of orally administered small molecules is frequently addressed through empirical model adaptations such as dose-dependent bioavailability or clearance. While effective and appropriate in some cases, these approaches may not sufficiently address all underlying structural drivers of non-linearity and result in biased population predictions. Mechanistic descriptions of saturable processes have been proposed in the context of target-mediated drug disposition (TMDD) and related approximations [1–3]. While TMDD-based approaches are well established for biologics, their systematic application to small molecules exhibiting apparent distribution-driven non-linearity remains limited. The objective of this work was to apply a stepwise modeling strategy to diagnose and characterize non-linear PK of small molecules, and to explore structural model components incorporating both pre-systemic and systemic saturable processes.
Objectives:
● Apply a stepwise structural modeling strategy to diagnose non-linear PK
● Compare empirical dose-dependent parameter scaling with mechanistic representations of saturable processes
● Propose a model structure incorporating both pre-systemic and systemic saturation
Methods: Early drug development PK data from an investigational orally administered small molecule at five single dose levels were analysed using nonlinear mixed-effects modeling. All models were fitted to individual concentration–time data in NONMEM using FOCE estimation. To illustrate dose-dependent behaviour and model performance, dose-normalized mean concentration profiles were evaluated across dose levels.
Model development followed three steps. First, a two-compartment model with linear clearance (CL) was evaluated using an empirical approach with dose-dependent relative bioavailability (F). Second, post-systemic non-linearity was explored using mechanistic and semi-mechanistic structures, including TMDD models implemented with different approximations (Michaelis–Menten, quasi-equilibrium, quasi–steady-state / quasi–distribution, and Wagner) [1–3], as well as systemic saturable binding models described using a Bmax function. Finally, a model incorporating two saturable binding compartments, representing non-linear processes prior to entry into the central circulation (pre-systemic) and after systemic entry, was evaluated without empirical bioavailability scaling. The concept of saturable distribution processes has been described previously [4].
Model performance was assessed by agreement between observation and prediction, systematic dose-dependent bias in population predictions (PRED), dose dependency in random effects, and changes in objective function value (OFV).
Results: The empirical model incorporating dose-dependent F improved individual fits but showed systematic population-level bias, with over-prediction at lower doses and under-prediction at higher doses. Pronounced dose-dependent bias in random effects for CL and V indicated unresolved non-linearity.
Introduction of post-systemic saturable processes improved model fit (ΔOFV>100 compared to empirical parameter scaling alone) and reduced bias; however, residual dose dependency remained across dose levels.
The final model incorporating both pre-systemic and systemic saturable binding compartments resulted in a substantial improvement in fit (ΔOFV>100 compared to the models with combined empirical bioavailability scaling and post-systemic saturable process), eliminated systematic prediction bias, and resolved dose-dependent trends in random effects. The estimated capacity of the pre-systemic saturable process (Bmax_presys = 0.8 mg) was of similar magnitude to that of the systemic saturable process (Bmax_sys = 1.0 mg), indicating that non-linear processes prior to systemic circulation contributed substantially to the observed non-linear PK alongside systemic saturation. This structure adequately described the observed non-linear PK without the need for empirical parameter scaling.
Conclusion: This work illustrates how stepwise structural model exploration can support the diagnosis of non-linear PK in orally administered small molecules. When empirical dose-dependent adaptations and post-systemic saturable processes alone are insufficient, a model incorporating both pre-systemic and systemic saturable binding processes may be considered as a practical alternative to empirical parameter scaling in clinical development, provided that such saturable processes are mechanistically plausible for the compound under investigation. This framework supports the characterization of non-linear PK and informed model selection in clinical development.
References:
References:
[1] Mager DE, Jusko WJ. General pharmacokinetic model for drugs exhibiting target-mediated drug disposition. J Pharmacokinet Pharmacodyn. 2001;28:507–532.
[2] Mager DE, Krzyzanski W. Quasi-equilibrium pharmacokinetic model for drugs exhibiting target-mediated drug disposition. Pharm Res. 2005;22:1589–1596.
[3] Gibiansky L, Gibiansky E. Target-mediated drug disposition model: relationships with Michaelis–Menten model. J Pharmacokinet Pharmacodyn. 2009;36:25–52.
[4] Peletier LA, de Winter W. Impact of saturable distribution on pharmacokinetic models. J Pharmacokinet Pharmacodyn. 2017;44:161–173.
Reference: PAGE 34 (2026) Abstr 12203 [www.page-meeting.org/?abstract=12203]
Poster: Methodology - New Modelling Approaches